1274
s.laTUEL
I. EPSTEINAND
fragile macromolecuIar structures has been kept to below 40%. Indeed, experimental uncertainties are such that probably only about half of the differences observed are due to distortions produced in macromolecules. I n addition to achieving a precision in the determination of average dimensions of asymmetric macromolecules nearly comparable with other methods, the electron microscope procedure provides much more detail of the distribution of lengths than heretofore available. Although this too may suffer some from residual distortional effects, it offers direct access to that type of information that is most difficult to obtain by other methods. It is therefore of interest to note that the comparisons made in this study have shown general agreement not only in average values but in breadths of distributions as well : the ichthyocol being quite narrow and the other two being close to the most probable distribution as predicted. With this check on the reliability of the distribution determination as well, the new electron microscope method appears ready to contribute to a number of problems where detailed knowledge of size-distribution of asymmetric macromolecules is desired. I t seems desirable to add a note concerning a
CONTRIBUTION
KO. 1466 FROM
THE
s. J. SINGER
Vol.
so
quite different method of measuring average molecular weights of macromolecules with the electron microscope that is applicable, independent of shape and capable of much higher accuracy. This method which involves counting the number ratio of standard to unknown particles in a given volume has been adapted to electron microscopy by \&?Iliams and Backus" to determine the molecular weight of tonlato bushy stunt virus (mol. wt. 10,000,000). This is potentially the most accurate method €or the determination of molecular weights by electron microscopy, but there are technical difficulties in the application of the method to particles with molecular weights in the range l o 4 to lo6, although such particles can be seen clearly by the methods used in the present study. These difficulties are mainly : (1) the production of droplets small enough to be entirely included in the field of the microscope a t suitable magnifications, ( 2 ) availability of standard particles not too d i f ferent in size from that of the unknowns and ( 3 ) adsorption or interaction between standard and unknown species.
-
(11) R (1949)
C Wllliams and R C Backus. TIIIS J O U R N A L 71, 4050
CAMBRIDGE, MASSACHUSETT~
STERLING CHEMISTRY LABORATORY, YALEUNIVERSITY]
Physical-Chemical Studies of Soluble Antigen-Antibody Complexes. IX. The Influence of pH on the Association of a Divalent Hapten and Antibody'.' B Y SAMUEL
I. EPSTEIN3 AND
s. J. SINGER
RECEIVED AUGUST5 , 1957 A quantitative study has been made of the light scattering from mixtures of a divalent benzenearsonic acid hapten and anti-benzenearsonic acid (anti-R) antibody (Ab), over a mide range of pH. Analysis of t h e data a t alkaline pH reveals t h a t a single ionizable group with pK 9.9 i. 0.2 is present in each Ab site, which must be ionized for association to occur appreciably. Ultracentrifuge experiments with these mixtures are consistent with this interpretation. Acetylation of anti-R Ab, under conditions specific for the blocking of the E-NHs+group of lysine, inactivates it, while an excess of a specific antigen, benzenearsonic acid-azo-bovine -,-globulin (RBG), protects anti-R Ab against this inactivation. Taken together, these experiments denionstrate t h a t a single t - S I & + group of lysine is critically present in each anti-R .Ab site and therefore t h a t a salt-linkage between this group and the negatively charged hapten is a n essential feature of the specific binding. The light scattering data,also suggest t h a t a t acid pI3, anions, especially phosphate, effectively compete with benzenearsonic acid hapten for combination with the Ab sites and that this competition obscures a n y effect t h a t the titration of the benzenearsonate ion might be expected t o have on the association of hapten and rlh. The parallelism in the properties of this hapteriXb system, and t h a t containing the natural protein antigen bovine serum albumin and its specific -\b is discussed.
It has long been assumed that an important feature of the bonds formed in many antigen (Ag)antibody (Ab) systems is the coulombic interaction of one or more pairs of oppositely charged groups located in the specific reactive regions of the two molecules. This assumption has derived partly from the fact that those haptens, such as benzenearsonic acid, benzenesulfonic acid, etc., which on being coupled to proteins, are most effective in eliciting antibody production, bear electric charges a t physiological PH. The inference has been drawn that oppositely charged groups are present in the (1) These studies were supported in part by grants from the National Microbiological Institute, United States Public Health Service, a n d from the Rockefeller Foundation. Parts of this paper were presented before the meeting of the American Chemical Society in Atlantic City, September, 1956. (2) T h e previous paper in this series is F. A. Pepe and S. J. Singer, THIS JOURNAL, 78, 4583 (1956). (3) United States Public Health Service Postdoctoral Fellow of t h e National Microbiological Institute, 1954-1956. Dept. of Chemistry, Tufts University, Medford, Mass.
Ab sites directed against these haptens. Some support for this inference is provided by the experiments of Nisonoff and P r e ~ s m a nwho , ~ found that the substitution on a hapten of an uncharged nitro group for a negatively-charged carboxyl group resulted in a marked reduction of the interaction of the hapten with antibodies directed t o the p-(p-azophenylazo)-benzoate ion. While this demonstrates that the negative charge on the hapten is critical, it does not prove, however, that positively charged groups occur in the antibody sitee5 One of the ob( 4 ) A. h-isonoff and D. Pressman, T K I SJ O U R N A L , 79, IC16 (1957). See also D. Pressman, A. L. Grossberg, L. H. Pence and L. P.~ulini: i b i d . , 68, 250 (1946). (5) I t is possible t o interpret this result without requiring a complementary charge in the Ab site. If the negatively charged hapten polarizes, or "binds," water molecules, then on formation of the hapten-Ah bond this water might be released, irrespective of the nature of the groups in the Ah site. This would result in a considerable entropy increase favoring t h e reaction. O n the other hand, the uncharged hapten, binding less water to it, would react less favorably with Ab.
March 20, 1958
INFLUENCE OF
p H ON iZSSOCIATION OF A DIVALENT HAPTEN AND ANTIBODY
jects of the present study is to provide direct evidence concerning the presence of charged groups in the reactive sites of Ab directed to a charged hapten. Another object of this investigation is connected with recent quantitative studies of the effect of acid pH on the equilibria in protein Ag-Ab systems.”? These studies, involving the antigens bovine serum albumin and ovalbumin and their specific rabbit antibodies, have been interpreted to indicate that a single ionized carboxyl group is present in each Ag-Ab bond. The studies reported in this paper, utilizing a simpler system, provide a critical test of this interpretation. We have performed light scattering and ultracentrifuge measurements of the effect of p H on mixtures of the divalent hapten (Hp), terephthalanilide-p,p’-diarsonic acid (T) and rabbit antibodies directed against benzenearsonic acid (anti-R Ab). Here the hapten is negatively charged near neutral pH. Because of the small molecular weight of T compared to anti-R Ab, the reaction of the two results in the formation of aggregates which are essentially dimers, trimers, etc , of Ab, from the point of view of light scattering and ultracentrifuge measurements. The light scattering behavior of this system a t p H 8 previously has been determined.8 We now find that the effect of alkaline p H in causing dissociation of Hp-Ab bonds is quantitatively accounted for by the hypothesis that there is a single titratable group, with a pK corresponding t o the e-ammonium group of a lysine residue, critically present in each Ab site. In the acid pH region, where the dissociation of Hp-Ab bonds might be expected to involve the titration of the benzenearsonate ions, anomalous behavior is observed which appears to be due in part t o competitive buffer ion binding to Ab. Materials and Methods Hapten-Antibody Mixtures.-The anti-R Ab was taken from the pool designated “54” in an earlier study.8 After isolation in July, 1954, this Ab was kept a t 5” for two months in p H 8.0, p 0.15 borate buffer; i t was then suspended in 40% saturated (NH4)&h. Most of the measurements at pH above 8.2 described here were made in M a y , 1955. Within probable experimental error, there was no change in the light scattering from a typical T-anti-R mixture, at p H 8, compared with t h a t found with freshly-prepared material, given in ref. 8. I n the ultracentrifuge, 88 ( ~ t 3 ) of 7~ the protein sedimented at the rate of y-globulin; the remainder (see pattern in Fig. 4 below) sedimented somewhat more rapidly and is likely an Ab dimer. Since no ultracentrifuge pattern of the fresh material had been obtained, the origin of the faster-sedimenting component is uncertain. The measurements at p H below 8.2 were made in February and April, 1956. Sgain, a t p H 8, t h e scattering from a T-anti-R mixture agreed well with earlier values. Moreover, there was no change in the ultracentrifuge pattern. Finally, storage did not change the precipitability of the anti-R with “R8’-Resorcinol,” a trivalent H p which was used in the purification of the pool.* The stored anti-R sediment was dissolved and dialyzed t o prepare anti-R solutions in 0.15 h’NaCl. The T used in this work was from the same lot as in the study at P H 8. The concentration of anti-R as well as T solutions was determined spectrophotometrically.8 T-anti-R mixtures were then made u p b y weight. (6) S. J. Singer and D. H. Campbell, THISJOURNAL, 7 7 , 3504 (1955).
(7) S. J. Singer, L. Eggman a n d D. H. Campbell, ibid., 77, 4855
1275
Light Scattering Measurements .-Several stock solutions which were about 0.1% in protein and had a T-anti-R mole ratio (r) of about 1 were prepared. T h e r-value was calculated on the assumption, used throughout these Hp-Ab studies, t h a t all protein in the purified solutions used is active anti-R, at suitable p H . Such a stock solution was clarified and the protein concentration in it determined as described earlier.8 Portions with a volume of 2.5 ml. were diluted by weight in a 10 mm.-square cell with 1 ml. of a series of buffers which had been clarified by filtration, The cell was stoppered with a clean, tight-fitting Teflon cap and by slowly inverting a few times, the contents were mixed without introducing appreciable optical impurity. T h e scattering measured just after mixing, which required about a minute, lay within less than 370 of t h a t found after 1 t o 4 hr., except as noted below, and presumably represents an equilibrium state. T h e final pH always was determined since it generally differed from t h a t of the buffer used. Most PH above 8 were attained with either 0.2 d l H3B03, 0.15 M in NaC1, titrated with 0.15 Jf XaOH; or p 0.15 h’a2HPO4-Xa8POa mixtures; or 0.001 to 0.01 ;If KaOH solutions brought to p 0.15 with NaC1. Other diluents, used chiefly for attaining acid p H , were p 0.15 acetate, arsenate, maleate, phosphate, phthalate and tris-(hydroxymethy1)-aminomethane (THAM)-HCl buffers, or solutions 0.00015 to 0.003 124 in HCl, brought t o g 0315 with NaCl. I n a few cases, more concentrated diluents were used t o attain a final p of about 1. Sodium salts nere used throughout. The light scattering measurements were made in a Brice-Speiser absolute photometer. Light of wave length 436 mp was used and dn/dc of Ab was taken as 0.200 a t all P H , although i t was determined8 only at pH 8. Ultracentrifuge Measurements.--For examination in the Spinco Model E ultracentrifuge, a stock mixture in 0.15 M NaCl was made in which the initial protein concentration was 1.03Y0 and r was 1.06. The mixture became cloudy and flocculation soon occurred, but after the mixture stood overnight and the precipitate was removed, the supernatant solution now remained clear for a t least t s o weeks. Such flocculation was never observed after making up the dilute mixtures used in light scattering. The optical density of the supernatant of this solution, a t 280 nip, was 95(+2)Y0 of the value one would have expected if no precipitation had occurred. If we assume that the precipitate contains equal moles of T and anti-R, this number gives, very nearly, the percentage of anti-R which remained in solution. An almost identical mixture was made up (initial protein concn., 1.05y0; initial r, 0.99) to demonstrate reversibility of the pH effects. In this case, only 80(+_3)70 of the anti-R, estimated in this way, remained in solution. The PH of these T-anti-R solutions was not measured. Although the T solution had been adjusted t o pH S in the course of dissolving the free acid and the anti-R had been dialyzed against saline adjusted to p H 7 with L-aOH, absorption of CO, b y these unbuffered solutions probably resulted in a p.H below 7. Since t h e mixtures were unbuffered, it is possible t h a t their p H differed and t h a t this affected the extent of precipitation. Portions of the stable supernatant were diluted by weight using tmo parts T-anti-R solution to one of buffer, with some of the buffers described earlier, and the pH was then noted. The mixtures were run at 50i40 r . p . m . at temperatures near 25”. Area measurements were made essentially as described elsewhere. Other Methods.-Electrophoretic analyses were performed in a Perkin-Elmer Model 35 Tiselius apparatus. p H was determined with a Beckman Model G meter, with correction for the sodium error. .4 Beckman Model DU or a Cary spectrophotometer was used for ultraviolet absorption measurements.
Experimental Results and Discussion Light Scattering Results at pH above 8.2.Two preliminary observations are pertinent. First, the slope of a light scattering plot of KclRgo Z J S . concentration up to 0 . 0 8 2 ~ 0for anti-R in pH 10.73, p 0.15 phosphate buffer is zero, within probable ex-
(1955).
(8) S. I. Epstein, P. Doty and W. C. Boyd, ibid., 7 8 , 3306 (1956).
(9) S. J. Singer and D. H. Campbell, i b i d , 7 7 , 3499 (1955).
SAMUEL I. EPSTEINA N D S. J. SINGER
127(i
VOl. 80
perimental error (*3% in Kc/RSO),that is, the Typical equilibria in basic solution may then be virial coefficient B is zero, just as it is a t CH 8.0. written At any intermediate pH, B will also be zero, since Hp Ab+2 HpAb+'; 4Ki (1) only electrostatic repulsions might reasonably be Abi'= H p S b ; 2Ki (2) Hp expected t o affect B , and these must be less imHp HpAb+' HpAbHp; Ki (3) portant than a t p H 10.73. We can thus plausibly r2b" + H+ J_ Ab+*; K H / ~ (4 take Mw,the weight-average aggregate weight, in TAb0 H+ Ab+'; ~ K H ( 5) anti-R mixtures throughout this range as RSo/Kc. Secondly, the effect of pH on the extent of associaHpAbo + H + HpAb+'; K n ( 6) tion is essentially reversible, a t least up to pH The symbol a t the right is the equilibrium con10.99, which bounds the region of interest. This stant characterizing each reaction ; the superscript will be discussed in more detail further on. indicates the number, j , of Ab sites per species which Most of the light scattering observations a t p H are both free and charged. Hp, in this context, above 8.2 were made on mixtures derived from a represents a T molecule, and Ab, an anti-R molecommon stock solution and are listed in Table I. cule. All other equilibria are characterized, in the In the third column, we record the values of KcIR90, chosen model, by the same intrinsic constants, determined by light scattering, and in the fourth, K i and KA, with a suitable entropy factor. Light Rgo/160000Kc that is, since B is zero, iVlw/M~, scattering observations, we shall see, yield p , the where MA is the antibody molecular weight. The equilibrium fraction of reacted haptenic groups. molecular weight of anti-R determined from light Let To and A , be the total concentration of hapscattering a t pH 10.73 was 162000 10000, which tenic groups and anti-R sites, respectively. If we agrees with the value obtained a t pH 8.0 for the were unaware that some anti-R sites were disfresh material.'O charged and inactive, we would then compute an In order to evaluate these data, we may consider apparent association constant a t this juncture the following simple model for the (hapten-antibody bonds) T-anti-R system, a model which is essentially iden- Knpp (free haptenic groups) (free Ab sites, charged or not) tical to that used by Singer and Campbell with pro(7 ) tein Ag-Ab system^.^^^ It is assumed that: (a) each of the two reactive sites of anti-R contains a single group which must be positively charged in order for the bond to form; (b) these groups are K a p p is related to K i through KH,for (hapten-antibody bonds) characterized by a single intrinsic acid association K . = '-(free, charged haptenic groups) (free, charged Ab sites) constant,ll KH; (c) the negatively charged hap(9) tenic groups undergo no significant changes in the pH range where the groups in the anti-R sites are titrated; and (d) non-specific electrostatic repul1 sive forces between Ab molecules do not appreci(1 ably influence the Hp-Ab association.ll It is also assumed, as in the studies a t pH 8, that all anti-R since (free, charged Ab sites) sites are identical and equally accessible to T. (11)
+ + + +
e
+")
KH E (uncharged Ab sites) (H+)
(IO) If t h e rapidly-sedimenting component noted in t h e ultracentxifuge is dimer, then we estimate from t h e areas under t h e two peaks t h a t MA is 179000 i 5000 (weight average). We use t h e usually accepted value of 160000 because it agrees with t h e light scattering value: if t h e latter is in error, i t is likely on account of factors which enter into Rw/Kc for a T-anti-R mixture in t h e same way, so t h a t t h e ratio Rso/160000Kc is correct. Moreover. traces of t h e Ratresorcinol used t o purify t h e anti-R were present in t h e pool and m a y cause aggregation only a t the much higher concentrations used in the ultracentrifuge. (11) More correctly, account should be taken of two factors: (1) the electrostatic effect of t h e other charges of t h e Ab molecule on the apparent K H of t h e critical group; and (2) t h e non-specific electrostatic repulsion of t h e Ab molecules on t h e apparent Ki of t h e I I p - l b association. If we use the functions of 'p and 6 t o rxpress these effects respectively, equation 12 should be written
- .
log
L+-
-
(12'1)
Blip
+o and 4 depend on ionic strength; on t h e location of t h e Ab sites on t h e highly asymmetric yglobulin molecule; and on t h e net charge on t h e Ab molecule a n d hence on pH. At present i t is not possible t o give correct explicit form t o these functions in this system, b u t i t is clear t h a t their effects are opposed and tend t o cancel. An increase in alkaline pH and in net negative charge Z on t h e Ab molecule makes i t more favorable for the critical group t o accept a proton a n d must therefore increase 'p. On t h e other hand, a n increase in Z increases t h e mutual repulsion of Ab molecule a n d hence decreases iI.. Furthermore, if t h e critical groups are located near t h e poles of t h e elongated Ab molecules, both 'p and 4 mill certainly be close t o unity a t p 0.15 and a t all values of Z which are encountered. Q'e assume, therefore, t h a t o$ r 1 and varies b y nu more than about 157, between $11 9 : i t i d 1 0 ,i in t h i q a v , t r n i
If we recast (10) to obtain a form more appropriate for treating data obtained over a range of pH, we have"
-.
log
lk -
Thus for the model chosen, a plot of log [l ' K a p p
- l / K i ]versus pH must have a slope of + l . O , and
from the intercept K H may be computed. If two such critical groups were present in each Ab site, the relation equivalent to equation 12 would be quadratic in pH, with a coefficient of 2.0 for the term linear in pH. If we express Mw in terms of the concentration of aggregates, the relation of Mwto p follows. Neglecting the weight of free and bound T +2
+1
(Hpn-iXhk)
+ J = o (Hpdlb:,) 4(Hpn+~-ibd)(nMd2
INFLUENCE O F pH
March 20, 1958
ON ilSSOCIATION O F A
Combining (13) with (1) - (6) and equations for analogous equilibria, and noting that
DIVALENT HAPTEN.4ND
- 4.5
I
I
I
1377
XNTIBODY
1
I
-
and that (HP) =
we obtain M,v
-
Ail
1 1
1 2'1 - P)?
+ 4K,2(Xb'2)(Hp)
- 4KL2(Ab+')(Hp)
- 1
+ pzv
1-
p7/
(14)
In these equations, r is defined as To'Ao. The final relation is identical with the one obtained for mixtures in which all Ab sites are charged.8 Application of equations 14 and 8 to the data of Table I results in the values of K a p p listed in column 5 . I n Fig. 1, these results are plotted in accordance with (12), using K , = 4.0 X lo5 l./mole, the value found a t p H 8.23, where essentially all of the anti-R sites are presumably active. The points are shown fitted with a line of slope 1.15 although a line with slope 1 fits nearly as well. This result indicates that the data are indeed compatible within probable experimental error with the simplified model discussed above. Furthermore, from equation 12 we obtain the result that log K H = 9.9 ri: 0.2, that is pKa of the group in the Ab site which is H the being discharged is 9.9, where K, = ~ / K is acid ionization constant for the group. The only acid groups in proteins with p K a = 9.9 are cysteine -SH, tyrosine -OH, and the t-NH3+ of lysine.12 The guanidinium group of arginine and the CY NH3+ group are normally characterized by pKa values of about 14 and 8. respectively. Of these posTABLE I DEPENDEXCE O F LIGHT S C A T T E R I N G F R O M T-AXTI-R TURES O N PH ABOVE PH 8 23 Conditions: pH
7 =
Protein concn., % '
RfIX-
1.15, 0.15 buffers, borate below p H 10, phogphate above PH 10. 106Kc/Rso
M,/MA
Rangea of lO-fiKSpD
4.0 8.23 0.067 3 . 0 0 =t0 . 0 9 2 . 0 8 zt 0.06 ,067 3.19 i .10 1 . 9 6 i .06 3 . 2 -3.7 8.99 9.48 ,067 3.37 i .10 1 . 8 5 i ,136 2 . 7 -3.2 9.68 .070 3 . 4 9 i. .10 1 . 7 9 i. . 0 5 2 . 4 -2.8 ,067 3 . 7 2 i .11 1 . 6 8 =t . 0 5 2 . 1 -2.5 9.90 10.22 ,067 4 . 7 0 i .14 1.33 =t .04 0.96-1.2 10.30 ,069 4.80 i .14 1.30 5~ .04 .84-1.1 15 1 . 2 7 rt .04 10.42 ,069 4.92 =t .75-0.99 10.54 ,067 5.25 f .16 1 . 1 9 f .04 .54- .78 .40- .58 10.68 ,067 5 . 5 5 i . 1 7 1 . 1 3 i .03 These are the probable extreme values of 10-6Kappif we assign a probable experimental error of +37, t o 106Kc/ Rooand hence, also, t o M,,,/MA. The possible presence of a small amount of dimer in the anti-R pool was ignored in computing these values: all the protein was assumed to be monomeric as well as divalent anti-R.
9.0
9.5
10.0
10.5
PH. Fig. 1.-The
variation of the apparent equilibrium constant data of Table I.
drogen-bonded to the benzenearsonate ion. This hydrogen bond would be unstable either upon the ionization of the tyrosine -OH or upon protonation of the benzenearsonate ion, and hence the dissociation behavior of such a linkage would be the same as that of an ion-pair interaction. The data up to this point cannot distinguish between these two possibilities, but the results of chemical modification studies to be discussed later on indicate that it is a lysine residue which is present in the Ab site. A complicating feature of the T-anti-R system is that arsonic acids are dibasic, and the second dissociation occurs a t alkaline pH. I n order to determine to what extent the alkaline dissociation of Hp-Ab bonds is affected by this ionization, the second dissociation constant of T was determined spectrophotometrically. The method was based on the fact that the ultraviolet absorption of T a t the 280 mp maximum drops about 12% when the PH changes from 6.5 to 9.5, without noticeable change in A., Three-ml. portions of p 0.15 NaC1 were diluted by solution containing 70 1g.T "1. weight with 1-ml. portions of p 0.15 borate or phosphate buffers. The optical density of each solution was measured against a blank consisting of a solution brought to p H 11.2, where all the arsonic acid groups are doubly ionized If we write, schematically, for the dissociation of the second H + From a single haptenic group As-
-15-
+ II+, K ,
115)
we have
sible acid groups, only that of lysine has the correct
pKa and bears a positive charge in the p H range where Hp-Ab association is maximal. It is possible, however, that instead of a positively charged lysine residue a tyrosine -OH in the Ab site is hy(12) E. J. Cohn and J. T. Edsall, "Proteins, Amino Acids and Peptides." Reinhold Publ. Carp., N e w York, IT. Y . , 1943, p. 446.
The (As-) a t any $ J His proportional to the observed difference in optical density, m; the (As=) a t the same PH is proportional to m m a x - m, where mmax is the maximum difference in optical density observed, that is, for the case where the solution contains T almost all of whose groups are singly-
charged (pH 6.5).
Thus
m log mmax - nz
=z
PIC,
- PH
(17
We assume in this derivation that the contribution to the optical density of a T molecule with one singly ionized group and one doubly ionized group is exactly midway between that of a T molecule with both groups singly ionized and one with both groups doubly ionized. The symmetry of the plot of v z vs. PH, Fig. 2, is consistent with this assumption.
-2t 7
8
9
IO
P H. Fig. 3.--Variation of optical density of T with p H : comparison of data of Fig. 2 with model for the effect of the second ionization of haptenic groups on optical density.
I 4
I
I
5
6
7
I
,
8
9
I1
,IO
II
PH. Fig. 2.-The variation of optical density of T with pH; m is the difference between the optical density of a solution a n d t h a t of a solution a t pH 11.2; concentration of T, 52 pg./ml.; p 0.15 borate or phosphate buffers; uncertainty in m, *0.010.
I n Fig. 3, the data of Fig. 2 are plotted according to (17). The line which fits the data has a slope of -1, in accordance with (17); pK2 = 8.2 f 0.1. This finding is consistent with the values found by Pressman and Brown13 in an extensive study of aromatic arsonic acids. The fact that pKz is much less than PKH of the Hp-Ab dissociation demonstrates that the second ionization of the arsonic acid is not involved in the dissociation. The data of Fig. 1 begin a t PH 9.0, by which point about of the arsonic acid groups are already doubly ionized. Indeed, it is surprising that the second ionization has an apparently negligible effect on the Hp-Ab reaction One might expect it to increase the coulombic interaction of Hp and Ab sites and hence to increase the association and the turbidity of T-anti-R mixtures; on the contrary, a 10% drop in turbidity occurs between p H 8.2 and 9 2 , which can be entirely accounted for in our model by the partial titration of the group in the A4bsite with pK 9.9. For the same reason, the use of K i calculated from a turbidity a t p H 8.2, where half the haptenic ( 1 3 ) J? Pressman and D. H . Brown, THISJOURNAL, 65, 540 (1943).
groups are still singly charged, is justified, even though essentially all the haptenic groups are doubly charged in the p H range where the largest fraction of the turbidity change occurs. Ultracentrifuge Results.--For the purpose of testing the PH dependence of aggregation by another method, we have also subjected T-anti-R mixtures to ultracentrifugation. I t is not possible to resolve by electrophoresis anti-K and its aggregates with T, which are essentially anti-I1 polymers. The sedimentation patterns are presented in Figs. 4 and 5 , along with that of anti-R itself. Three not well-resolved peaks are distinguishable, a t most, in the T-anti-R mixtures. These peaks have sr0 of approximately 6.5-7, 9-10, and 12-13 svedbergs; we shall refer to these peaks as m, d and t, respectively. Presumably m represents the species Ab, AbHp and AbHp,, since it sediments a t the rate of y-globulin. The sedimentation rates of peaks d arid t are consistent with those which would be anticipated for anti-I< “dimer” and anti-R ‘‘trimerj’’the sum of species containing two anti-R InoIecules and three anti-R molecules, respectively. l 4 (14) These 520 values are t h e observed sedimentation rates, corrected only for t h e temperature dependence of t8e solvent viscosity, this being t h e principal correction npplicable in these dilute buffers, excepting concentration dependence. A t infinite dilution, s& will exfor t h e molecule is ceed these values. Taking szo for anti-R as ?, +,of., found t o be 1.5 from t h e Svedberg equation. If t h e departure of f/fo f r o m unity is ascribed solely t o asymmetry, and we assume t h r prolate ellipsoid model, the Ab molecule has a n axial ratio of 9. If side-by-side aggregation occurs and we assume t h a t the aggregate is a n ellipsoid as long as the monomer. and whose diameter is chosen so that the volume of t h e aggregate equals the degree of polymerization multiplied by t h e volume of monomer, we find t h a t $0 is 12.6 and 17.3, for t h e dimer and trimer, respectively. T h e same method of calculation, using a complementary assumption, yields s& of 8.8 and 9.6, respectively, f o r end-to-end dimer a n d trimer. T h e observed sedimentation rates of peaks d a n d t lie between these limiting estimates.
March 20, 1958
INFLUENCE OF pH ON ASSOCIATIONOF
A
DNALENTHAPTENAND ANTIBODY
1279
If we interpret the patterns in this way, those a t The area under an n-mer peak will be proportional neutral pH are most interestingly consistent with to the weight concentration of total n-mer, that is, the earlier light-scattering studies of this system8: to %(totaln-mer). Thus, where k is a constant reaggregation is maximal when 7 is about 1, and in ex- lating area under peak m to molar concentration of cess of either reactant, aggregation is diminished. total monomer (M), and At& is the total area under A t alkaline pH above 10, aggregation diminishes all peaks abruptly, again reflecting ligbt scattering behavior. = k ( ~+ ) 2 k p + ( ~ )+ 3 k ( p * , ) * ( ~+. ) . . . . . . . (19) The sedimentation pattern can provide a picto- Then rial proof that the effect of pH is reversible under the conditions used. A T a n t i - R solution was 3% 6 = (1 - $+)2 (20) Am Atot brought to pH 10.99 by the slow addition of pH 11.8, p 0.15 phosphate buffer. After 0.5 hr., it was where A , is the area under the peak m. We have neutralized to pH 7.55 by adding pH 4.85, A, 0.15 used this relation to obtain p and hence Kappfrom acetate buffer. The sedimentation pattern (Fig. the data listed in Table 11; the results are recorded 6) is nearly superposable on that of a mixture iden- in column 6. The values are consistent within the tical except that the acid and alkaline buffers were large experimental error with those estimated from first mixed and then added to the T-anti-R soh- light scattering, column 7, although they are unition. Thus appreciable denaturation of anti-R formly lower. This may be explained by the prewas absent in our experiments. vious finding*that the equilibrium constant apparUpon attempting to obtain a more quantitative comparison of the ultracentrifugal and light scattering results, we encounter the problem of the poor resolution of the peaks in the sedimentation patterns. This is due, in part, to the small difference in sedimentation rates of the peaks but may also be affected by 1.51 the occurrence of re-equilibration reactions as an adjustment to loss of larger aggregates in various parts of the cell. For example since Ab, AbHp and HpAbHp sediment a t the same rate, reactions of the type Ab HpAbHp HpAbHpAb may occur as dimer sediments out of the region of the monomer peak. In contrast, the free Ag concentration can be ob1.50 tained unequivocally in systems like bovine serum albumin (BSA)-anti-BSA, since the sedimentation of aggregates leaves a region in e the cell which contains only BSA, so that there e 4 is no possibility of further reaction. While such reequilibration reactions, if they were of an appropriate rate, would be expected to diminish progressively the area ascribed to monomer in successive patterns of a single ex1.65 1.0 periment, this was not observed within the errors of analysis. Because of the poor resolution, corrections for the Ogston-Johnston anomalies, and for the fast-sedimenting component present in the anti-R itself, were not considered warranted. With these reservations on the accuracy of the data, we may proceed to interpret the 4.1 1.71 results of Table 11: which lists the areas under each peak in the T-anti-R patterns of Figs. 5 Fig. 4.-Ultmcentrifu~e diagrams of T-anti-R mixtures with and 6. If peak m indeed represents anti-R different ,-values, in a 0.15 borate buffers. pH near 8, compared monomer, then the area under it permits with that of anti-R (top row). The patterns at three different evaluation of Kapp. We note first that accord- times in a single experiment are presented, from right to left. ing to our model, (total Ab n 1-mer) is chronologically. Sedimentation proceeds to the left. Peaks given by (p27) (total Ab n-mer), where par- labelled m. d, and t refer to monomer, dimer. trimer, as disentheses indicate concentration in moles/liter cussed in text. where appropriate. For example ently decreased with increasing total concentration, and the solutions examined in the ultracentrifuge were about ten times more concentrated than those studied by liaht scatterina. There is a sharp drop +2 +I in K,,, whiih coincideswith the abrupt change observed by light scattering, between pH 10.11 j-0
+
-f
-
se $
+
SAMUEL I. EPSTEIN AND S. J. SINGER
1280
pH 8.27 10 11 10 64 Fig. 5.-Ultracentrifuge diagramsdemonstrating the effect of alkaline pH onT-anti-Rassociation, pO.15 buffers. Mole ratio Hp/Ab = 1.06, Ab concn. = 0.65%. Single patterns from each experiment were observed at roughly comparable times of sedimentation. Sedimentation proceeds to the left. Peaks labelled m, d. and t refer to monomer dimer and trimer, as discussed in text.
and 10.64, and which reflects the shift in the proportions of anti-R in the two dominant peaks (Fig. 5). TABLE I1 SEDIMENTATION PATTERNS OF T-ANTI-RMIXTURES
Vol. 80
plotted Rpo/160000Kc, that is, if the virial coefficient is zero, M,/MA, against PH. T o guarantee that the virial coefficient is indeed zero for all the solvents in which the mixtures were examined, would require measurements of the molecular weight of antibody in all buffers used over the PH range 3-8. Actually, one would anticipate that the virial coefficient, a t least a t these low concentrations and moderate ionic strengths, would have only a negligibly small effect on the turbidity. The few data that were obtained on 0.06% solutions of anti-R itself supported this expectation. Thus, 106Kc/Rpowas 6.02, 5.75, 6.55 and 6.34 in p H 8.2 borate, pH 6.7 phosphate, PH 3.9 acetate and p H 3.0 phthalate buffers, respectively, all a t p 0.15. The data suggest a small decrease in the turbidity a t low pH,but the average deviation of the four results from the mean value 6.17 (expected value for B = 0 and MA = 160000, 6.25) is about 570, which is only a little larrrer than the probable experimental error.
p 0.15 buffers; I = 1.06 for the first three solutions; r = 0.99 for the others. Buffer composition-UC data: first two solutions. u 0.05 nhosohate. Y 0.10 NaC1: third solution, u 0.05 b o k (&tai& NaCl), II 0.10 NLCI; last two solutions, p 0.015 acetate, p0.04 phosphate, pO.095 NaCL.. The uncertainty in the LS (light scattering) values for K-,, is about *lo%; in the UC (ultracentrifuge) values, because of the D O O ~resolution. 2 ~ 5 0."' 7 ~ . % % Total area vnder peak 10-8 K .,
pH
Protein
m
d
t
uc
LS*
67 I 1 0 33 & 10 0.2 0.5 0 . 9 1.4 31 + 10 69 i 10 8.27 .G5 16 & 5 5 7 i 10 27 i 10 2 . 2 4 . 0 7.55 .50 35 i 10 52 & 10 13 5 0 . 9 2.1 7.50 .50 3 0 i 10 55 i 10 1 5 + 5 1 . 2 2 . 1 Values estimated from data of Fig. 6, which pertain to solutions of slightly different buffer composition. 10.64 10.11
0.65 .65
There are, however, some inconsistencies in the ultracentrifuge diagrams. If we obtain p2r from equation 20, we can compute the expected distribution of the remaining area among the other peaks. I n the case of the pattern a t PH 8.27, the fraction of the total area which should appear as dimer and trimer, presumably under the d and t peaks, is 19 and 17%, respectively; the other 48Y0 of the anti-R should be involved in larger aggregates. On the other hand, the observed areas under the d and t peaks are much larger than expected. The explanation for this discrepancy is not clear a t present. It may be that the assignment of the d and t peaks to dimer and trimer species only, is an oversimplification. Furthermore, re-equilibration reactions may seriously distort the area distribution in these patterns. Light Scattering from T-anti-R Mixtures below $8.2.-The €I scattering from T-anti-R mixtures a t PH below 8.2 was observed, and a few more mixtnres a t PH above 8.2 were examined as a check on the previous result. The data obtained are shown in Fig. 7, together with all the measnrements from which Fig. 1 was derived." We (15) T h e turbidity 01 the mixtures which were brought to p 1.2 with sodium phosphate was 10% lower, 4 hr. alter mixing, than the turbidity just after midog. In these e_*, no observation was made 1 hr. after miring, (IO it is not known whether the drop occurred p d u ally or during the 6nt hour.
(a) Fig. 6.-The reversibility of the effect of alkaline pH on Tanti-R association. Pattern (a) is of a stock solution exposed to pH 10.99. and returned to pH 7.55; (b) stock solution diluted t o have same concentration and bu5er composition as (a). TheHp/Abmoleratio is0.99, final Ab concn. 0.50~06.The buffer contains phosphate ions, see text.
As was indicated earlier, i t was originally expected that the acid dissociation of Hp-Ab bonds would be determined by the titration of the negatively charged benzenearsonate ion. If Kxh is the reciprocal of the first ionization constant of the haptenic group and we adopt a model entirely analogous to the one previously discussed for the alkaline Hp-Ab dissociation, it follows that log
(-Kspp 1 -) 'K,
=
log
s h - pH K.
(21)
This equation is identical in form to that derived by Singer and Campbell.'.' I n this context If,. = (hipten- ontihody honds) (free Iuptctiic groups, charged UT not)(free Ab sites) ~~
From the value of K i = 4 X lo6,and the assumed value of KHh = lo', which is at the upper limit of values found for typical benzenearsonic acids,13 we may calculate, from equations 21, 8 and 14, the values of Roo/160000Kcas a function of acid p H which are Dredicted bv this model. The calculations were 'made for two different sets of r-values and anti-R concentrations: r = 1.15, anti-R concn. = 0'067%; and = ''0°, anti-R conm' = o'060%; to correspond closely to the mixtures actually stud-
'
INFLUENCE OF p H
March 20, 1958
ON
ASSOCIATION OF
A
DIVALENT HAPTENAND ANTIBODY
1281
striking that the turbidity a t a given acid PH is lowest in phosphate buffer and decreases with increasing concentration of phosphate. These effects observed by light scattering are apparent in certain of the ultracentrifuge patterns as well. Thus the pattern a t $H 7.5 (Fig. 6) in a buffer containing phosphate shows a large increase in the m component, a t the expense of the d and t components, compared with the pattern a t pH 8.2 in borate buffer. It has long been known that arsenate itself inhibits the precipitation of anti-R Ab by specific substances, l7 clearly because of a competition for Ab sites. Our one light scattering measurement with arsenate present in the buffer indicates t h a t a t the same $H and molar concentration phosphate and arsenate are essentially equally effective in reducing the turbidity of T-anti-R mixtures. We conclude that phosphate can also react with anti-R Ab sites, which is not surprising in view of the close structural relationship between phosphate and arsenate ions. This effect of phosphate in causing the dissociation of Hp-Ab bonds I
0
1.00. I
I
I
I
I
I
>-I
desc.
Fig. 8.-The demonstration of t h e protection of anti-R by an excess of RBG, against inactivation by acetylation. Descending electrophoresis patterns of: top, acetylated R B G a n t i - R mixtures; bottom, a mixture of separately acetylated R B G and rabbit 7-globulin. Both experiments in barbital buffer, p H 8.6, p 0.1, under identical conditions. The fast peak in each diagram is due to acetylated RBG, and the arrow indicates t h e starting position and direction of migration.
takes precedence over any similar effect that might be produced by the titration of the hapten groups. Between pH 8 and 5.5, the turbidity of mixtures in phosphate buffer drops more than 30%, t o a value nearly equal to that of a solution containing antiR alone; yet T remains essentially fully charged under these conditions. The light scattering data suggest, moreover, that in the acid PH region where Ab molecules bear a net positive charge, maleate, acetate and chloride ions, in order of decreasing effectiveness, can also com(17) K. Landsteiner, Biochem. Z . , 104, 280 (1920). See also F. Haurowitz and F. Breinl. 2. physiol. Chem., 2i4, 111 (1933).
1282
SAMYEL I. EPSTEIXAND S.J. SINGER
Pete for the anti-R site. The influence of pH, in a given buffer system, might therefore be very complicated, involving a t least the following factors: (a) the net positive charge on anti-R increases as the p H is lowered below 7, which should increase the binding of buffer anions to the specific, as well as numerous other, sites on the anti-R molecule; (b) T hapten is an anion, and a t acid p H may bind non-specifically t o anti-R a t other than the reactive sites of the Ab molecule, thus increasing the extent of Hp-Ab association by effectively increasing the Ab “valence” beyond 2 ; and (c) finally, the titration of the charged groups on the hapten. The factors (a) and (b) should oppose one another; as the p H is progressively decreased, the former should decrease, the latter increase, the apparent extent of Hp-Ab association, and the resultant variation might be complex indeed. This could account qualitatively for the observed variation of the turbidity of T-anti-R mixtures with acid pH. An attempt to treat this situation theoretically does not seem warranted a t the present time. At first sight it might seem unjustified to regard the light scattering data obtained a t alkaline pH as relatively free of complications because they conform to simple theoretical expectations, and those a t acid p H as anomalous because they do not, but our interpretation is supported by a number of internal consistencies. It is, in fact, logical that a t alkaline PH, where Ab molecules bear a net negative charge, competition of buffer anions for anti-R sites should be much less significant than a t acid pH, where Ab is positively charged. (Non-specific ion-binding by bovine y-globulin a t pH about 7 is very small.18) Therefore the alkaline p H studies might be expected to be freer of such complications. Furthermore, experimentally, the turbidity results a t alkaline p H are essentially the same in borate, phosphate or NaC1-NaOH solutions. In addition, the order of effectiveness of the buffer ions in reducing the extent of Hp-Ab dissociation a t a given acid pH is correlated roughly with the extent of their structural similarity to the benzenearsonate group, and hence their capacity to react with anti-R sites. I t follows from these arguments that considerable interest would attach to a similar study of a Hp-Ab system in which the Hp is positively charged. In this case, the alkaline pH range might yield anomalous results due to the possible competition of cations with the H p for the Ab site, while the acid p H range might simply reflect the titration of a negatively charged group in the Ab site. A question which this discussion provokes deserves comment: is the value of Ki obtained in earlier studies* in @ H 8 borate buffer affected by either buffer inhibition or noli-specific binding of T ? The most abundant anion in pH 8 borate buffer is chloride, since the pK of boric acid is about 9: the coincidence of the results in borate and THAM buffers presumably reflects this common property. From the results a t acid pH, it is plain that chloride ion is the weakest inhibitor among the ions used here. Thus, while an estimate of the effect on Ki cannot be given, it should be comparatively small. (18) I. M. Klotz, in t h e “Proteins,” ed. b y H. Neurath a n d IC. Bailey, Vol. I B , Academic Press, New York, N. Y.,1953, p. 787.
Vol. 80
The non-specific binding of T to anti-R should be negligible a t PH 8 in comparison with the specific binding, since it was found earlierg that data over a range of r-values were all compatible with an anti-R valence of two. However, while the effect is small, generally lying within probable experimental error, it is perhaps significant that the aggregation in mixtures in which r was about 3 persistently exceeded that expected from the data for mixtures with r about 1. Weak non-specific binding of T would indeed be favored by high T concentration, and the resultant increased possibility for aggregation would only become apparent a t high r values where aggregation by the completely specific mechanism is diminished because of T excess inhibition. I t should be noted finally that our results suggest that light scattering might be used effectively to determine absolute values of hapten inhibition constants in systems in homogeneous equilibrium. Chemical Modification Studies.-In order t o investigate further the nature of the groups present in anti-R Ab sites, some preliminary studies were made of the effect of acetylation of the Ab under conditions which are specific for the blocking of free amino g r o ~ p and s ~ which ~ ~ ~ do ~ not cause the reaction of tyrosine-OH groups. For these experiments the antigen employed was a heavily-coupled benzenearsonic acid-azo-bovine y-globulin (RBG) used in other studies.21 The anti-R was not pure i l b but was part of a preparation containing mostly inert y-globulin.22 Treatment of RBG with acetic anhydride a t 0’ in half-saturated sodium acetatelg permitted it to retain lOOyo of its capacity to precipitate anti-R Ab. This is as expected, since the acetylation reaction should have no effect on the benzenearsonic acid hapten. On the other hand, the anti-R y-globulin fraction treated in the same manner, with the resultant blocking of about 75% of its free amino groups,23 lost its capacity to precipitate any measurable amount of RHG, as determined 24 hr. after mixing the two protein preparations. The detailed methods used in these experiments are essentially thosc described previously. 2o The inactivation of anti-R by acetylation may be due either to a non-specific denaturation of the molecule consequent to the blocking of 75y0of its normally free amino groups or to the blocking of essential amino groups in the two reactive sites, the blocking of all the other amino groups being inconsequential. That a non-specific denaturation is not involved is shown by the fact that an excess of RBG protects anti-R from inactivation by acetylation, presumably due to the protection of the Ab sites by the Ag.24 This protection was deinonstrated as A precipitate of REG and (19) H . F r n e n k e l - C o n r a t , R
S R r i n and 11. I . i n t 177, 38.5 (1919). S J. Singer. Proi. YutL. A c n d . S c i . 1:. Z.. 41, 1041 (19.73) F. I’epr and S. J . Sinprr. t o be published. \I. C. Raker, D. 11. Campbell, S. I . Epstein a n d S . 1 Sinarr, T H I SJOUKN.IIL, 78, 312 ( 1 S X ) . This preparation 1s referred to :ts antiR - I in this paper. (23) As determined by \*an Slyke analyses: D. D. V a n Siyke, J . B i d . Chem., 83, 425 (1829). (34) D. Pressman and L. A. Sternberger, J . Immrcnoi., 66, 609
fhfn1.. (20) (21) (22)
(1951).
March 20,
1
3
~
~
OF
~HISTIDINE ~ ~ COMPLEXES ~ ~ OF ~ COBALT(I1) ~ € ~AND YCOBALT(II1)
anti-R was dissolved in excess RBG to give a mixture of soluble Ag-Ab aggregates. This mixture was acetylated and subsequently examined electrophoretically (Fig. 8, top). As a comparison, a mixture of separately acetylated RBG and rabbit -y-globulin was also examined (Fig. 8, bottom). It is evident that a peak due to Ag-Ab aggregates, of mobility intermediate between that of acetylated RBG and acetylated 7-globulin, is present in Fig. 8, top, indicating that the anti-R retained considerable activity after acetylation in the presence of RBG. Therefore, the inactivation of anti-R which takes place upon acetylation in the absence of RBG would appear to be due to the blocking of essential amino groups in the specific Ab sites. This information complements the light scattering results with T-anti-R mixtures at alkaline pH discussed earlier and we conclude that in each Ab site there is critically present a single e-NH3* group of a lysine residue, complementary to the benzenearsonate hapten. Comparison with Protein Ag-Ab Systems.In the-systems containing bovine serum albumin (BSA) and ovalbumin (OA) and their respective rabbit antibodies, quantitative studies have been made6.' in the ultracentrifuge of the effect of acid p H of the extent of Ag-Ab association. It was found that the data in both systems obeyed a re-
[ COXTRIBUTION
FROM THE
1283
lation equivalent to equation 21, with Kh % IO5, and i t was therefore concluded that a single ionized carboxyl group was critically present in each AgAb bond formed in these systems. The studies a t alkaline PH in the T-anti-R system provide another independent example of such behavior and lend considerable support to the conclusions reached in these earlier studies. The behavior of the BSA:anti-BSA system toward acetylation is also quite parallel to that of the RBG :anti-R system.20 BSA retains most (70y0) of its activity upon acetylation, but its specific Ab is inactivated under these conditions. Furthermore, an excess of BSA protects its Ab from inactivation by acetylation. We conclude, therefore, that the two apparently unrelated systems have a great deal in common: a single ion-pair is involved in each of the bonds of BSA :anti-BSA and of R : antiR , with one of the groups being an E-NH~Iin each Ab site. This parallelism is very likely reflected in the very similar thermodynamic parameters characterizing these system^.^^^^ Acknowledgments.-We thank Drs. W. C. Boyd and P. M. Doty for permission t o use the pool 54 anti-R and Drs. D. Pressman and D. H. Campbell for the gift of the necessary haptens. NEW HAVEN, CONN.
DEPARTMENT OF CHEMISTRY,
UNIVERSITY OF MICHIGAN]
The Polarography of Histidine Complexes of Cobalt (11) and Cobalt(II1) BY BRUNOJASELSKIS RECEIVED JULY 25, 1957 Unlike most cobalt(I1) complexes hihistidinatocobalt( 11) shows a n anodic wave in buffer solutions above pH 5.5; the potential of which becomes more negative with a n increasing pH until a limiting value of -0.25 v . us. S.C.E. is reached. The presence of the anodic wave is attributed to the uncharged bihistidinatocobalt( 11)species being oxidized in a reversible manner t o bihistidinatocobalt(II1) ion. The anodic wave for bihistidinatocobalt(I1) and the cathodic wave, the first wave for Co(II)(hi)2. The cobalt(I1) complexes of hisbihistidinatocobalt(II1) ion, are for the same couple Co(III)(hi)z+ f e tidine methyl ester and histamine show a similar anodic wave above pH 8, presumably due t o the oxidation of uncharged hydroxo complexes.
The polarographic reduction of cobalt(I1) in the presence of histidine has been reported It has been observed that histidine depressed the cobalt(I1) reduction maximum producing a similar maximum a t a more negative potential. In the present paper the polarographic behavior of bihistidinatocobalt(T1) and of bihistidinatocobalt(II1) ion is reported in greater detail. In particular the polarograms have been examined in the region between -0.1 v. and -0.40 v. vs. S.C.E. in various buffer solutions. The presence of a well-defined anodic wave for bihistidinatocobalt (11) in buffered solutions above p H 5.5 and the lack of an anodic wave for the cobalt(I1) complexes of histidine methyl ester, and histamine in buffer solutions below pH 8 is attributed to the oxidation of the uncharged complexes. This is substantiated by the studies of cobalt(I1) complexes of histidine, histidine methyl ester and histamine a t various p H ' s . (1) J. Sladek a n d M. Lipschutz, Coli. Ceeck. Chem. Cotnnt., 6, 487 (1934). (21 E. R. Roberts, Trans. F a r a d a y Soc., 37, 363 (1941).
Furthermore, the relationship between the first cathodic wave for the reduction of bihistidinatocobalt(II1) ion and the anodic wave for the oxidation of bihistidinatocobalt(I1) is established. Experimental Work Materials.-The crystalline bihistidinatocobalt(I1) was prepared by the reaction of 0.01 mole of cobalt(I1) hydroxide with 0.02 mole of pure histidine in 200 ml. of deaerated water. Subsequent vacuum evaporation yielded the crystalline complex having a n empirical formula ClrHleNsOICo.H20. The results of analysis of this crystalline complex are summarized in Table I. T h e bihistidinatocobalt(II1) aqueous solution was pre-
TABLE I SUMMARY OF ANALYSIS OF CRYSTALLINE BIHISTIDINATOCOBALT ( I I) Analysis
Percentage
of
found
Percentage calcd.
Co (total) Co(I1) N (Dumas) C H
15.35 15.24 22.02 36.81 4.68
15.33 15.33 21.82 37.40 4.67
